Journal of the American Statistical Association volume 80, issue 390, P393-397 1985 DOI: 10.1080/01621459.1985.10478129 View full text
|
|
Share
J. F. Lancaster, Dana Quade

Abstract: A nonparametric test for hypotheses concerning linear regression parameters is based on the signs of the residuals and their Kendall correlation with the independent variable. The smallsample distribution of the test statistic is tabulated, asymptotic approximations are derived, and the corresponding confidence region is presented. Furthermore, the Pitman asymptotic relative efficiency of the test is calculated with respect to the classical test based on least squares and with respect to the BrownMood test.